Studies on the wood corrosion on a historic glulam-construction

Wolfgang Rug1, Gunter Linke

2, Angelika Lißner

3

ABSTRACT: The exposure to chemically-aggressive media causes a structural modification of wooden members. The

related mechanisms are called wood corrosion. The subject of the studies was an approximately 100 year’s old

warehouse, which was executed as HETZER-construction and has been used as storage for potassium salt. The

assessment of the degree of modification and the evaluation of the strength reduction was the aim of the studies. Test on

the strength as well as chemical analyses have been carried out. The results showed a clearly visible modification of the

examined material due to the exposure to the salts. This was recognizable by the superficial macroscopic modifications

as well as by a reduction in the strength of the material in the periphery of the cross-section.

KEYWORDS: wood corrosion, chemically-aggressive media, Otto Hetzer, glulam, Dynstat-method

1 INTRODUCTION 123

The invention of the glued laminated timber by Otto

Hetzer in the early 20th

century allowed the building

wide-spanning constructions [2]. Amongst others halls in

the fertilizer and potassium salt industry were erected by

the use of this construction method, which is proven

until today by several preserved buildings (see also

www.otto-hetzer.de).

However, this field of application subjects the building

material wood to particular environmental conditions

[6]. For example, the production processes as well as

stored substances are releasing chemically-aggressive

media, which can cause irreversible structural

modifications of the wooden members.

The modification of the wood due to chemically-

aggressive media is not taken into consideration in the

static calculations according the currently valid version

of the Eurocode 5 [7]. Some suggestions concerning this

problem are listed in [6].

Though, the knowledge of the degree of the modification

is a significant requirement for static calculations

concerning the load bearing capacity.

2 WOOD CORROSION

Due to its chemical balance wood has a high resistance

against chemically-aggressive media. If the correct

environmental conditions are present, the wooden

members will be subjected to corrosion [6].

1 Prof. Dr.-Ing. Wolfgang Rug, Hochschule für nachhaltige

Entwicklung, Eberswalde, wolfgang.rug@hnee.de 2 Dipl.-Ing. (FH) Gunter Linke, Hochschule für nachhaltige

Entwicklung, Eberswalde, linke@holzbau-statik.de 3 Dipl.-Ing. (BA) Angelika Lißner, Ingenieurbüro Rug GmbH,

Wittenberge

The wood corrosion is a structural damage beginning on

the surface, which is caused by chemical and physical

reactions. These reactions are caused in particular by

strong acidic and strong alkaline media (pH ≤ 2

respectively pH ≥11).

If there are salts involved, the cell structure is destroyed

by crystallization processes. The lignin and the

hemicelluloses are degradated due to hydrolytic splitting.

The in the individual case present destructive mechanism

is dependent on the type of the impacting media. This

fact is a result of numerous investigations on timber

constructions additionally stressed due to chemically-

aggressive media.

The level of the destruction depends on the several

aspects. These aspects could also be called a corrosion

system, which is depicted in Figure 1.

Figure 1: schematic depiction of the wood corrosion

On a large amount of the so far observed constructions it

was determined that the destruction of the wood

structure is limited to the cross-section near the surface.

This destruction is visible by a greyish-brown

discolouration and a fibrous surface structure. In some

cases the separation of whole strips of wood along the

annual ring limits was observed. Furthermore, the

peripherical cross-section shows a reduction of the

strength as well as an increased density. In the inner

cross-section there is no such strength reduction.

Glued laminated timber has a higher resistance

concerning the wood corrosion as long as large, compact

members are used and no large shrinkage cracks are

appearing. Furthermore, the type of glue significantly

influences the resistance against chemically-aggressive

media.

3 SUBJECT AND AIM OF THE STUDY

The effect of the salts on the load bearing capacity of the

historic glued laminated timber has been examined in the

course of material tests [1]. The main focus laid on tests

with the Dynstat-method as well as on chemical analyses

concerning the salt content.

The results of this technical material study should be

used as basis for future investigations concerning the

structural stability of existing glued laminated timber

constructions which are additionally stressed by

chemically-aggressive media.

The subject of the study was a warehouse which was

erected in 1912 by the use of the HETZER-construction

method. The load-bearing structure of the warehouse

consisted of eleven parabolic trusses made from glued

laminated timber according to the patent DRP. 197773

[5]. The trusses (wood species: spruce) had a double-T

cross-section (see Figures 2 & 3).

Figure 2: view on the truss construction; the examined section is marked in red

After the demolition of the ware house in April 2010

several parts of the construction have been transferred to

the University of Applied Sciences, Eberswalde/

Germany (HNE Eberswalde) to carry out technical

studies on the wooden members.

Figure 3: sectional view A-A of the truss construction – 1: top

chord; 2: web; 3: solid wood bracing; 4: bottom chord (all dimensions in mm)

The sample material had clearly visible marks of

corrosion. The surface was greyish-brown discoloured

and fibrous. The discolouration reached a depth of 20-

25mm from the surface. Furthermore, the already

mentioned separation of wooden strips could also be

found. The Figures 4 and 5 are showing the macroscopic

appearance of the sample material exemplarily.

Figure 4: top: fibrous surface structure and greyish

discolouration as well as salt deposits on the bottom chord;

bottom: sectional view on a chord and the web (the borders of

the discolouration are marked in red)

Figure 5: discolouration of the peripherical cross-section,

shown on a bar-shaped sample from a chord lamella (the discoloured area is crosshatched in red)

4 RESEARCH METHODOLOGY

4.1 SAMPLE EXTRACTION

The extraction of the samples was limited to an area of a

length of approx. 1,50m near the support of the truss (see

Figure 2). The observed part of the truss has shown the

clearest corrosion marks of all the available truss

fragments. This can be explained by the direct contact

between the salt and the wooden members during the

service life of the warehouse.

Bar-shaped samples with a cross-section of b/h =

10/15mm and a length of ℓ = 65mm (web) respectively 205mm (chords) have been extracted from each lamella

of the chords and the web of the truss. The sample

extraction included every lamella of the observed truss

fragment. In total 60 bar-shaped samples have been

extracted – 30 samples from the cords and 30 samples

from the web.

These bar-shaped samples have been cut into

approximately 4mm thick disc-shaped Dynstat specimen

along the longitudinal direction of the sample (see Figure

6). In this way, it was possible to create depth profiles

which cover the whole length of the samples.

Figure 6: top: schematic depiction of the specimen cutting;

bottom: separation of a bar-shaped sample into disc-shaped

specimen at the example of a sample from a chord lamella

Each sample from the trusses web was cut into fourteen

disc-shaped specimens. The samples from the chord

lamellas were divided into 44 disc-shaped specimens

each. In total 1740 specimen were cut from the overall

60 samples as shown below.

Sample series „chord“:

30 samples with 44 specimen 1320 specimen

Sample series „web“:

30 samples with 14 specimen 420 specimen

4.2 STRENGTH TESTS WITH THE DYNSTAT-METHOD

The remaining strength of the material has been

determined by flexural tests with the Dynstat-method

according to DIN 53435:1983 [9] respectively. TGL 1-

204/8:1965 [10]. In the nowadays valid version of the

test specifications [9] the application of the Dynstat-

method is restricted for the test of plastic materials.

Studies [11] have proven the principally applicability of

this test method for technical studies on wood. This

applies especially for the material-saving determination

of the material properties and the development of

strength profiles along the width respectively the height

of the cross-section.

The specimens have been loaded with a continuous rate

of deformation until a failure occurred. The test setup is

depicted in Figure 7.

Figure 7: test setup of the Dynstat-method – 1: specimen; 2:

rotation axis; 3: bending support (r = 0,5mm); 4: bending blade (r = 0,5mm)

The test apparatus used is shown in the Figures 8 and 9.

Figure 8: Dynstat apparatus (type Dys-e 9303)

Figure 9: specimen in the test apparatus during the test

The flexural tests were accompanied by the

determination of the density according DIN 52182:1976

[12] and the wood moisture according DIN EN 13183-

1:2002 [13] by the determination of the dimensions and

the weight of the conditioned specimen (υ = 20°C and

φ = 65%) as well as in dry state.

4.3 CHEMICAL ANALYSIS OF THE SALT CONTENT

The salt content was determined by the use of the optical

atom emission spectroscopy with inductive coupled

plasma (ICP-OES). This analysis method is based on the

interaction between the elements contained in the

samples and the electromagnetic radiation emitted by

them.

By stimulating a vaporized sample solution, the outer

electrons of the contained elements are raised to a higher

energy level. The supplied energy is emitted during the

relapse of the electrons to their base energy level in the

form of light. The emitted light has a specific

wavelength and frequency which is characteristic for a

specific element (see [14]). By measuring the

wavelength and the intensity of the emitted light the

contained elements and their concentration can be

determined in comparison to given calibration data.

This chemical analysis was carried out exemplarily on

one sample from the web and the chords. The samples

were chosen according to their distinctive strength

profile. In total 58 specimens have been analyzed

chemically.

5 RESULTS AND EVALUATION

5.1 RESULTS OF THE PHYSICAL-MECHANICAL TEST

The flexural test have been evaluated according the

regulations in DIN 53435:1983 [9] and TGL 1-

204/8:1965 [10]. Accordingly the maximum flexural

stress equals the bending strength of defect-free wood.

However, a conversion of the determined strength values

on the strength of small defect-free specimen in

accordance with DIN 52186:1978 [12] (bar-shaped

specimen, test configuration is shown in Figure 9) with

the help of suitable factors is required.

Such factors for the conversion of the bending strength

and the density have been determined in the last years at

the HNE Eberswalde (see [11]).

Figure 10: test configuration for the flexural test on bar-

shaped, defect-free specimen according to DIN 52186:1978 [12]

The relation between the properties of Dynstat-specimen

and bar-shaped, defect-free specimen (wood species:

spruce) is specified in [11] as shown in the following

Equations (1) and (2):

- density:

Dynstat

DIN52186

ρρ =

0.92 (1)

- bending strength:

m,Dynstat

m,DIN52186

ff =

0.54 (2)

where ρDIN52186 = density of bar-shaped, defect-free

specimen in accordance with DIN 52186, ρDynstat =

density of Dynstat-specimen, fm,DIN52186 = bending

strength of bar-shaped, defect-free specimen in

accordance with DIN 52186, fm,Dynstat = bending strength

of Dynstat-specimen

The influence of the wood moisture on the bending

strength has also been considered by the application of a

conversion factor. It is specified in [15] that the bending

strength decreases by 4% if the wood moisture increases

by one percentage point. This is valid for the increase of

the wood moisture below the fibre saturation point. The

equivalent conversion factor can be determined as shown

in Equation (3):

fm, Glk = - 0.04 +1 (3)

Where kfm,ω = conversion factor for the influence of the

wood moisture on the bending strength, ωGl =

equilibrium wood moisture in [%] which attunes under

the predominant climate (here: ωGl =12%), ω = wood

moisture during the tests [%]

After the conversion of the test results a statistical

analysis concerning possible statistical outliers took

place (test according to GRUBBS, see [22]). In the result

of this analysis no statistical outliers could be

determined.

Furthermore, the distribution of the test results has been

analyzed by the help of the KOLMOGOROV-

SMIRNOV-Test (see [22]). It was determined that the

bending strength as well as the density are approximately

normal distributed. The actual distribution is positively

skewed (rate of skewness: density: 0.551; bending

strength: 0.582). This means that the greater amount of

test results is located below the mean value of the

assumed ideal normal distribution.

The Figures 11 and 12 are depicting the absolute

frequency as well as the assumed ideal normal

distribution of the density and the bending strength of all

specimens.

Figure 11: distribution of the density of all specimens

Figure 12: distribution of the bending strength of all

specimens

The results of the statistical analysis of the converted test

results are listed in Table 1 and Table 2.

Table 1: results of the statistical analysis of the converted test results – sample series „chord“

density ρ12 bending

strength fm

amount of specimen n 1320 Stk. 1320 Stk.

minimum xmin 0.31 g/cm³ 23.2 N/mm²

maximum xmax 0.67 g/cm³ 141.9 N/mm²

mean value xmean 0.47 g/cm³ 82.1 N/mm²

standard deviation sx 0.06 g/cm³ 17.1 N/mm²

variation coefficient vx 12.1% 20.8%

Table 2: results of the statistical analysis of the converted test results – sample series „web“

density ρ12 bending

strength fm

amount of specimen n 420 Stk. 420 Stk.

minimum xmin 0.37 g/cm³ 38.3 N/mm²

maximum xmax 0.68 g/cm³ 164.4 N/mm²

mean value xmean 0.50 g/cm³ 88.8 N/mm²

standard deviation sx 0.06 g/cm³ 23.7 N/mm²

variation coefficient vx 12.0% 26.7%

The evaluation of the test results was carried out as a

comparison with the normatively regulated material

properties of spruce according to DIN 68364:2003 [16].

In this standard the mean density is given with a value of

0.46g/cm³ and a variation coefficient of 9.7%. The mean

bending strength has a value of 80N/mm² and a variation

coefficient of 14.2%.

The comparison between the test results and the material

properties according to DIN 68364:2003 [16] has shown

that the density of the examined material is slightly

increased. The mean density is located in the top

variation range of the normatively regulated density

(sample series “chord”: 0.47 g/cm³; sample series “web”:

0.50 g/cm³). The observed increase can be attributed to

the salt deposit in the peripherical wood structure. The

comparison of the mean density of the examined

material and the density according to DIN 68364:2003

[16] is depicted in Figure 13.

Figure 13: comparison between the densities of the examined material with the density according DIN 68364:2003[16]

The bending strength of the examined material is also

higher than the bending strength according DIN

68364:2003 [16]. An equivocal explanation for this

observation could not be found on basis of the studies

that have been carried out. The comparison of the mean

bending strength of the examined material and the

bending strength according DIN 68364:2003 [16] is

depicted in Figure 14.

Figure 14: comparison between the bending strength of the

examined material with the bending strength according DIN 68364:2003[16]

The evaluation of the mean values of the bending

strength has shown that the examined material was

obviously not affected by the impact of the salts (see

Figure 14). However, considering the depth profile of the

bending strength it is clear that the strength in the

peripherical cross-section is reduced in comparison to

the inner cross-section (see Figure 15).

Figure 15: depth profile of the bending strength – sample G1-2 (sample series „chord“)

5.2 RESULTS OF THE CHEMICAL TEST

In the result of the chemical analysis mainly potassium

chloride, sodium chloride and magnesium chloride could

be determined to be deposed in the wood structure.

Furthermore, several sulphates in lower concentration

were detected. The concentration of these sulphates

equals the amount of sulphates which have to be

expected in long-term obstructed timber. Therefore, the

effect of the sulphates on the examined material’s

properties is neglected.

The total concentration of the detected chlorides in the

peripheral cross-section has a value of approximately

10% (sample series “web”) respectively 13% (sample

series “chord”). The concentration decreases towards the

inner cross-section. The analyzed sample from the web

has a salt concentration of approximately 4-5% in the

inner cross-section. The concentration of the inner cross-

section of the sample from the chord was approximately

1-2%.

To qualitatively determine the influence of the salt

deposition on the bending strength a linear correlation

and regression analysis has been carried out. This

analysis refers to the entire specimens which have been

chemically analyzed. The results are depicted in Figure

16.

Figure 16: depiction of the relation between the salt concentration and the bending strength

Figure 15 shows clearly that the bending strength

decreases with the increasing salt concentration. This is

also proved by the negative correlation and regression

coefficients. It should be noted at this point that the

examined relation is only moderately expressed. The

correlation coefficient has only a value of -0.485. The

relatively small sample size of only 58 specimens can be

seen as the reason for this fact.

5.3 DETERMINATION OF THE CORROSION LAYER THICKNESS

The corrosion layer is the peripherical part of the cross-

section which shows a clearly visible respectively

measureable alteration due to the influence of the

chemically-aggressive media. These alterations are

manifested – as already described above – as a

macroscopic modification (discolouration, fibrous

surface structure) as well as a reduction of the material’s

strength.

The thickness of this corrosion layer can be determined

graphically on basis of the results of the technical

material studies which have been carried out. General

rules are listed in [17] which are derived from studies on

timber construction in the potash industry. These rules

are listed below.

1) The bending strength in a distance of 14mm from the surface is approximately constant. The mean

value is depicted as a straight line in this area.

2) The bending strength beyond this intact inner cross -section (core area) decreases steeply.

3) The thickness of the corrosion layer results from the intersection of the steeply decreasing strength in

the peripherical cross-section and the 75%-line of

the mean bending strength in the intact inner cross-

section (core area) plus the half of the specimen’s

thickness.

This method was applied exemplarily on the test results

of on sample of each sample series. These are the same

samples which were chemically analyzed.

For the sample from the sample series „chord“, the

thickness of the corrosion layer equals 13.5mm. The

sample from the sample series “web” the thickness had a

value of 11mm.

The Figure 17 depicts exemplarily the graphically

determined thickness of the corrosion layer of sample

G1-2 in comparison to the determined salt concentration.

Figure 17: graphical determination of the corrosion layer’s thickness according [17] (sample G1-2)

The Figure 17 reveals that the bending strength in the

assumed intact inner cross-section is partially lower than

in the damaged peripherical cross-section. The salt

concentration in the assumed intact inner cross-section is

relatively high with values up to 8%.

The in [17] described method has been derived from test

on small, defect-free specimen (bar-shaped specimen,

see Figure 9). Therefore, the unconditional applicability

of this method on the bending strength of Dynstat-

specimen is questionable.

Thus, a second graphical method for the determination

of the corrosion layer’s thickness was applied. The

second method uses the 5-percentile value of the bending

strength of defect-free wood according to DIN

68364:2003 [16] as reference. Then, the corrosion layer

is the part of the cross-section in which the test results

are lower than the reference value.

With the help of this consideration a corrosion layer’s

thickness of 5mm (sample series “web”) respectively

10mm (sample series “chord”) could be determined. The

Figure 18 depicts exemplarily the graphically determined

thickness of the corrosion layer of sample G1-2 in

comparison to the determined salt concentration.

The corrosion layer’s thickness which was determined in

comparison to the 5-percentile value of the bending

strength of defect-free wood according to DIN

68364:2003 [16] is seen as the decisive thickness.

Figure 18: graphical determination of the corrosion layer’s thickness (sample G1-2)

The main reason for this consideration is the nowadays

valid safety concept for the design of timber

constructions. This concept uses 5-percentile values as

the characteristic values for the design. Another reason is

– as described above – that the in [17] described method

can only be applied on results of the Dynstat-method

with restrictions.

Therefore, the thickness of the circumferential corrosion

layer of the examined material is declared with values of

5mm on the web respectively 10mm on the chords (see

also Figure 19).

Figure 19: thickness of the corrosion layer

5.4 COMPARATIVE CALCULATIONS CONCERNING THE STRUCTURAL

STABILITY

The knowledge of the thickness of the corrosion layer

creates the basis for the analysis of the structural stability

of a timber construction which is affected by the

influence of chemically aggressive media. A proposal for

the design of such constructions can be found in [6].

According to [6] there are two ways to statically

calculate such constructions. Both methods are similar to

the hot-state design according to Eurocode 5 [18]. On the

one hand, it is possible to calculate with the reduced

material strength due to the influence of the chemically-

aggressive media (method 1). On the other hand, the

cross-section can be reduced by the thickness of the

circumferential corrosion layer (method 2).

The comparative calculation concerning the structural

stability has been carried out by the use of both methods.

The results of the physical-mechanical and chemical

tests created the basis of the calculation.

Method 1: calculation with a reduced material strength

The modification factor to take account of the influence

of the chemically-aggressive media kmod,aM was selected

according to the results of the chemical tests.

Method 2: calculation with a reduced cross-section

The cross-section of the glulam trusses has been reduced

by a value of 5mm. This equals the thickness of the

corrosion layer which was determined for the web of the

trusses.

The results of the comparative calculation revealed that

the static utilization rate of the glulam trusses under

consideration of a reduced material strength (method 1,

kmod,aM = 0.95) equals the utilization rate under

consideration of a circumferential corrosion layer

(method 2, d = 5mm).

The results of the proposed design methods given in [6]

can be proved by the results of the technical studies

which were carried out. Therefore, these design methods

are adequate for the prognostic calculation of the

structural stability of timber structures under

consideration of the influence of chemically-aggressive

media.

5.5 DETERMINATION OF THE CHARACTERISTIC MATERIAL

PROPERTIES

Another initial value for the calculation of the structural

stability of a timber construction which is affected by

chemically-aggressive media is the assignment of the

affected members to the officially approved system of

strength classes for glued laminated timber members.

This is achieved by the determination of the

characteristic material values.

In this case the characteristic values were determined

according EN 384:2010 [18]. The test results of all tested

specimen (sample series “chord” and “web”) have been

divided into two new groups. The first group included

the entire specimen from the undamaged cross-section

(intact inner cross -section). The second group was

formed by the specimen from the corroded peripherical

cross-section. The advantage of such a partition is that as

the actual load bearing capacity of the undamaged cross-

section can be estimated more accurately.

The characteristic values of the strength properties are

determined according EN 384:2010 [18], chapter 5.4 as

shown below (see Equation 4).

k 05 S Vf = f k k (4)

where kf = characteristic value of the strength in

[N/mm²], 05f = mean value of the 5-percentile values

which are weighted in proportion to the amount of

specimen in [N/mm²], Sk = correction factor to consider

the amount of samples and specimen according to EN

384:2010 [19], Figure 1, Vk = correction factor to

consider the low variability of the 5-percintile values of

machine-graded timber in comparison to visually graded

timber

The 5-percentile value f05 is determined as shown in

Equation 5:

05 x 95f = f - s t (5)

where 05f = 5-percentile value of the strength in

[N/mm²], f = mean value of the strength in [N/mm²],

xs = standard deviation of the strength in [N/mm²], 95t

= t-factor of the Student-distribution for 1 - = 95 %

The characteristic value of the density is determined

according EN 384:2010 [18], chapter 8 as shown below

(see Equation 6).

05,j j

k

j

ρ nρ =

n (6)

where kρ = characteristic value of the density in

[kg/m³], 05, jρ = 5-percentile value of the density of a

sample in [kg/m³], jn = amount of specimen in one

sample

The 5-percentile value ρ05 is determined as shown in Equation 7:

05ρ = ρ -1,65 s (7)

where ρ = mean value of the density in [kg/m³], s =

standard deviation of the density in [kg/m³]

A characteristic value of the bending strength of

25.67N/mm² in the corroded peripherical cross-section

respectively 45.40N/mm² in the intact cross-section was

derived from the test results. The characteristic density

equals a value of 413.81kg/m³ in the corroded

peripherical cross-section respectively 376.23kg/m³ in

the intact inner cross-section.

These results are again proving the reduction of the

material’s strength in the peripherical cross-section in

comparison to the intact inner cross-section. The density

in the peripherical cross-section is increased. This fact is

caused by the deposition of salt in the peripherical cross-

section as already mentioned above.

The characteristic material properties of timber are

nowadays regulated in the EN 338:2010 [20], Table 1.

The comparison of the characteristic values of the test

results with the characteristic values of the strength

classes in EN 338:2010 [20] allows an assignment of the

material in the corroded peripherical cross-section to the

strength class C24. The material in the intact inner cross-

section can be assigned to the strength class C27

although the characteristic bending strength is much

higher than listed in EN 338: 2010 [20], Table 1 (see

also Table 3).

Table 3: possible assignment of the examined material to the strength classes of timber according EN 338:2010 [20]

C2

4

per

iph

eric

al

cro

ss-s

ecti

on

C2

7

inn

er

cro

ss-s

ecti

on

bending

strength

fm,k [N/mm²]

24 25.67 27 45.40

density

ρk [kg/m³] 350 413.81 370 376.23

Following the regulations of EN 14080:2013 [21], Table

1 it is possible to assign lamellas of the strength class

C24 to the T-class T14. Lamellas of the strength class

C27 can be assigned to the T-class T16.

Since the extraction of the samples included every

lamella in the cross-section the assignment mentioned

above can be assumed for the whole cross-section.

However, this assignment could only be seen as

orientation since the flexural modulus of elasticity is also

decisive in order to assign timber to a strength class.

Furthermore, the characteristic values of the test material

are derived from test on defect-free wood. The

transmission of the results to full-scale timber members

is only possible with the help of an adequate regression

function. At the moment there is only inadequate

knowledge for wood affected by chemically-aggressive

media.

On basis of the orienting assignment of the examined

material to the strength and T-classes of timber a

valuation on the load bearing capacity of the glulam

timber members is now possible.

According to EN 14080:2013 [21], Table 3 glulam

timber which is entirely made of lamellas of the T-class

T14 can be assigned to the strength class GL24h

according to EN 14080:2013 [21], Table 5. If the entire

lamellas are assigned to the T-class T16 the glulam

timber could be assigned to the strength class GL26h.

These possible assignments are only valid for modern

glued laminated timber. Whether this is also valid for

historic glulam members is questionable at the moment.

The main causes are the lack of knowledge about the

historically used glues as well as the fact that historic

glulam members were produced without finger joints.

6 CONCLUSIONS

The results of this study are revealing that the examined

material was clearly altered due to the salt exposure in

spite of its high resistance against the effect of

chemically-aggressive media.

The chemical analysis has shown that the salts are not

only deposed in the peripherical cross-section but also in

its core. The concentration was measured with a value of

10-13% in the peripherical cross-section respectively 1-

5% in the core.

The influence of the chemically-aggressive media is

amongst others clearly visible by the macroscopic

alteration of the wood structure near the surface. The

surface structure of the examined material was fibrous

and showed a greyish-brown discolouration. This

discolouration reached a depth of 20-25mm.

The results of the material test have proven the

macroscopic alteration of the examined material.

Although, there is no significant reduction of the

strength in the inner cross-section there is a significant

reduction of the strength in the peripherical cross-

section. The density decreases from the surface towards

the inner cross-section. This refers to the analyzed salt

concentration.

The thickness of the circumferential corrosion layer was

determined on basis of the examined material’s bending

strength. Two methods were used – firstly the method

according to [17] and secondly a comparison between

the test results and the 5-percentile value of the

normatively regulated bending strength according to

DIN 68364:2003 [16]. In the result a thickness of 5mm

on the web respectively 10mm on the chords was

determined.

A comparative calculation according to the method

described in [6] showed that the prognostic calculated

load factor (method 1 - calculation with a reduced

strength) approximates the actual load factor (method 2 -

calculation with a reduced cross-section).

The influence of the chemically-aggressive media on the

material’s properties is not taken into consideration in

the currently valid timber construction rules so far [7].

However, this study has shown that this influence has to

be taken into consideration on existing timber structures.

The results of this study led to further investigations on

the influence of the chemically-aggressive media on the

load bearing capacity of historic glued laminated timber

constructions. This includes the determination of the

strength of glulam and solid wood member in full-scale

as well as a study on the glued joints. These studies will

be mentioned in future publications.

ACKNOWLEDGMENT This study was financed by private donations. The

authors would like to thank the donors for the generous

support.

REFERENCES [1] Rug, W., Linke, G.: Untersuchungen zur

Holzkorrosion an historischem Brettschichtholz.

Bautechnik 90 (2013), Heft 10, S. 642-650

[2] Rug, W.: Innovation im Holzbau – Die Hetzerbauweise (Teil 1). Bautechnik 71 (1994), Heft

4, S. 213–219.

[3] Rug, W.: Innovation im Holzbau – Die Hetzerbauweise (Teil 2). Bautechnik 72 (1995), Heft

4, S. 231–241.

[4] Rug, W.: 100 Jahre Hetzer-Patent. Bautechnik 83 (2006), Heft 8, S. 533–540.

[5] DRP. 197773: Gebogener Holz-Bauteil für vereinigte Dach-Pfosten und -Sparren (1906)

[6] Rug, W.; Lißner, A.: Untersuchungen zur Festigkeit und Tragfähigkeit von Holz unter dem Einfluss

durch aggressive Medien. Bautechnik 88 (2011),

Heft 3, S. 177–188.

[7] DIN EN 1995-1-1:2010 – Eurocode 5: Bemessung und Konstruktion von Holzbauten – Teil 1-1:

Allgemeines – Allgemeine Regeln und Regeln für

den Hochbau.

[8] Linke, G.: Untersuchungen zur Holzkorrosion bei historischem Brettschichtholz. Hochschule für

nachhaltige Entwicklung (FH), Eberswalde 2011

(unveröffentlicht – diploma thesis)

[9] DIN 53435:1983 – Prüfung von Kunststoffen – Biegeversuch und Schlagbiegeversuch an Dynstat-

Probekörpern

[10] TGL 1-204/8:1965 – Prüfung von Holz – Bestimmung der Biegefestigkeit und

Schlagzähigkeit an kleinen proben (Dynstatprüfung)

[11] Rug, W.; Eichbaum, G.; Linke, G.: Vergleichende Festigkeitsuntersuchungen an Holz mit dem

Dynstat-Verfahren. In: Holztechnologie 52 (2011) 6,

S. 34-40

[12] DIN 52182:1976 – Prüfung von Holz – Bestimmung der Rohdichte

[13] DIN EN 13183-1:2002 – Feuchtegehalt eines Stückes Schnittholz – Teil 1: Bestimmung durch

Darrverfahren

[14] Gey, M. H.: Instrumentelle Analytik und Bioanalytik – Biosubstanzen, Trennmethoden,

Strukturanalytik, Applikation. Berlin: Springer

Verlag, 2008

[15] Niemz, P.: Physik des Holzes und der Holzwerkstoffe. Leinfelden-Echterdingen: DRW-

Verlag, 1993

[16] DIN 68364:2003 – Kennwerte von Holzarten – Rohdichte, Elastizitätsmodul und Festigkeiten

[17] Erler, K.: Bauzustandsanalyse und Beurteilung der Tragfähigkeit von Holzkonstruktionen unter

besonderer Berücksichtigung der Korrosion von

Holz. Technische Hochschule Wismar.

Habilitationsschrift, Wismar: 1988

[18] DIN EN 1995-1-2:2010 – Eurocode 5: Bemessung und Konstruktion von Holzbauten – Teil 1-2:

Allgemeine Regeln – Tragwerksbemessung für den

Brandfall

[19] DIN EN 384:2010 – Bauholz für tragende Zwecke – Bestimmung charakteristischer Werte für

mechanische Eigenschaften und Rohdichte

[20] DIN EN 338:2010 – Bauholz für tragende Zwecke – Festigkeitsklassen

[21] DIN EN 14080:2013 – Holzbauwerke – Brettschichtholz und Balkenschichtholz –

Anforderungen

[22] John, B.: Statistische Verfahren für Technische Meßreihen – Arbeitsbuch für den Ingenieur. Carl

Hanser Verlag, München – Wien. 1979

[23] Schwar, A.: Mazeration als Erscheinungsform chemischer Korrosion an verbautem Holz –

Auslösende Faktoren, Mechanismen, In: Bautechnik

77 (2000), S. 478-483